derivative for GELU (#1160)

* derivative for GELU

* add tests
This commit is contained in:
KGrewal1
2023-10-23 20:23:45 +01:00
committed by GitHub
parent eae94a451b
commit 807e3f9f52
2 changed files with 22 additions and 1 deletions

View File

@ -471,7 +471,15 @@ impl Tensor {
Op::Unary(_, UnaryOp::Round) => {
Err(Error::BackwardNotSupported { op: "round" })?
}
Op::Unary(_, UnaryOp::Gelu) => Err(Error::BackwardNotSupported { op: "gelu" })?,
Op::Unary(arg, UnaryOp::Gelu) => {
let sum_grad = grads.or_insert(arg)?;
let cube = arg.powf(3.)?;
let tanh = (0.0356774 * &cube + (0.797885 * arg)?)?.tanh()?;
let gelu_grad = (((0.5 * &tanh)?
+ (0.0535161 * cube + (0.398942 * arg)?)? * (1. - tanh.powf(2.)?))?
+ 0.5)?;
*sum_grad = sum_grad.add(&(&grad * gelu_grad)?)?
}
Op::Unary(_, UnaryOp::Erf) => Err(Error::BackwardNotSupported { op: "erf" })?,
Op::Unary(_, UnaryOp::GeluErf) => {
Err(Error::BackwardNotSupported { op: "gelu-erf" })?